AWS Auto Scale Alarm Trigger Failure

Question

Have configured an auto scaling group and a launch configuration that will use a scaling policy bound to a target group metric (RequestCountPerTarget, specifically).

However, it is failing to trigger the scaling event.

Auto Scaling Group

• The auto scaling group is configured with a proper and existing launch configuration, where there is not an active limit on instances.
• The maximum instance count is higher than the desired count.
• The "Activity History" tab shows zero events fired.

CloudWatch Alarm

• Creation of the auto scaling group and scaling policy did create an automatic alarm.
• The alarm did update state to ALARM, and trigger an automatic action.

Automatic Alarm Trigger Action

{
  "error": null,
  "actionState": "Succeeded",
  "notificationResource": "arn:aws:autoscaling:ap-southeast-1:0000:scalingPolicy:0000:autoScalingGroupName/WebScalingGroupV1:policyName/Scale API Requests",
  "stateUpdateTimestamp": 1519937155623,
  "publishedMessage": null
}

I have completely removed this group and configuration to recreate it -- as initially, I had spaces in the configuration and scaling group name; I am having the same issues.

Answer 1

After speaking with Manu, whom is an AWS support engineer, it was determined that the root cause preventing auto scaling actions to trigger was due to a configuration conflict whereby the Minimum and Desired instances were set to zero.

Amazon auto scaling groups will NOT trigger based on target tracking when the current capacity (Desired) is at zero, and apparently will also not give any clear indication of the fault via an error log.

Reference Target Tracking Considerations at https://docs.aws.amazon.com/autoscaling/ec2/userguide/as-scaling-target-tracking.html#target-tracking-considerations

A target tracking scaling policy does not scale your Auto Scaling group when its current capacity is 0, because target tracking scales proportionally to the current capacity.